Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Currently Trying To Figure Out How To Extend Clearml'S Automagical Reporting To Joeynmt.

Currently trying to figure out how to extend clearML's automagical reporting to JoeyNMT.
https://github.com/joeynmt/joeynmt/blob/master/joey_demo.ipynb is a Colab notebook demoing JoeyNMT. I added the pip install clearml and clearml-init and from clearml import Task and task = Task.init(project_name="project name", task_name="Esperanto_Bert_2") lines to the beginning. !python -m joeynmt train transformer_epo_eng_bpe4000_reload.yaml is the line that starts the training That line outputs lots of information like epoch number, etc. It also creates a tensorboard folder. However, the epoch number, validation loss, and so forth don't seem to be caught. Is there a way to specify "pay attention to this tensorboard folder"? /content/models/epo_eng_bpe4000_transformer/tensorboard is the path to the tensorboard folderEdit for future users: you have to do the training in the same python process where clearml was imported. So something like:
from joeynmt.training import train train("transformer_epo_eng_bpe4000.yaml")will work

Posted 2 years ago
Votes Newest

Answers 6

Hi SmallDeer34
ClearML automagical logging will work on the current python process. But in your example yyour Bash is running another python script (that has nothing to do with the original notebook), hence clearml automagic is not aware of it (i.e. it cannot "patch" the tensorboard calls).
In order to make it work.
you should do something like:
from joeynmt import train train.main(...)Or something similar 🙂
Make sense ?

Posted 2 years ago

It seems to create a folder and put things into it, I was hoping to just observe the tensorboard folder

Posted 2 years ago

Oh, of course, that makes total sense

Posted 2 years ago

Yup! That works.
from joeynmt.training import train train("transformer_epo_eng_bpe4000.yaml")And it's tracking stuff successfully. Nice

Posted 2 years ago


Posted 2 years ago
6 Answers
2 years ago
one year ago